I happen to have a copy of that one...It depends on many factors, some of which are discussed in this famous paper by Toole & Olive.
I happen to have a copy of that one...It depends on many factors, some of which are discussed in this famous paper by Toole & Olive.
Could you correct to stay in a 1dB window?I take this question as how flat is flat (enough). Using the LS50's as an example, their listening window in the Soundstage measurements fits in a 3 db window and the PEQ filters I use flattens it to a 1.5 db window. I would say the difference is pretty subtle when comparing my filters vs No EQ so based on that I would say if you can get your listening window within a 2db window (with a speaker with good directivity) you're in good shape.
I take this question as how flat is flat (enough). Using the LS50's as an example, their listening window in the Soundstage measurements fits in a 3 db window and the PEQ filters I use flattens it to a 1.5 db window.
Could you correct to stay in a 1dB window?
How did you measure that? Can you post that measurement here?
Sure, I can fix a mic at my LP, do a single sweep, invert the response to create a filter and when measuring again I will have response within 1.5dB window - but it will sound awfull. Once you do a proper linearisation of the in-room response based on spatially averaged measurements it will sound good, but it won't be within 1.5dB window when you measure it again.
I could but I don't like to go crazy with EQ above the transition frequency, the LS50's mostly just need the region from 2-5k brought down a bit and a slight cut around 750, slight boost around 1500.
Right, it wouldn't work using the room curve, I only EQ above the transition frequency using anechoic measurements if possible. I brought in the listening window from Soundstage's measurement and corrected the response, the Klippel measurement would be fine too but the 5k peak isn't evident in any of my in-room measurements so I trust Soundstage more in this case.
Even if we ignore room effects there is a difference between samples - your speaeker will not have identical response as the speaker tested by Soundstage.
I've had 3 pair of LS50 now and all 6 speakers were pretty close to each other so I'm not sure that's actually a real problem or maybe LS50s have better quality control than cheaper speakers. Either way you're right that no method is perfect but it's the we've got so it's what I go with.
She may be flat, but man what a set of stems!#sorrynotsorry
A minidsp will probably run out of filters halfway through this method
My own personal criterion is that if the frequency response is within 3dB (i.e. +/-1.5dB) over the audio band depending of course on the LF capabilities of the loudspeaker, that's good enough. I've equalised my own to be +/- 1dB above 200Hz as anything closer than that was just too much bother.
S.
Thanks for your recipe.Every speaker designer I've ever talked to has assured me that their speakers measure flat. Flat can be defined in a multitude of ways, unless you are dealing with perfect constant directivity, a +-0db on-axis speaker will not be 'flat'. Currently regardless of who you purchase a pair of speakers from, or how much you paid, you can always do better correcting them yourself - if you know what you are doing.
Here are my 10 steps to achieve a "flat" speaker and room combo for one listening position:
1. All measurements for corrections are done windowed outside for treble on the tweeter axis
2. All bass measurements for corrections are done using 2 meter ground plane method outside - do not window this measurement down.
3. Splice the two measurements together and correct flat +-0.5dB to at least 50Hz
4. Remeasure off axis(0,5,10,15,30,45,60 degrees horizontal and 0,5,10,15,30 vertical) treble response outside with correction applied.
5. Average the results of the second round of measurements, any irregular dips or peaks that do not follow the slope in average treble should be adjusted minimally to within +-1.0dB of the natural directivity slope of the speaker, compromising the on-axis response for a smooth weighted power response.
-- You can stop at this point if you want a pure "flat speaker" and let the room do what it wants, otherwise continue --
6. Place corrected speaker in final position for best imaging, measure speaker with the mic pointed in listening position at ear level pointing at the speaker - do not window this measurement down.
7. Take 1:1 smoothing response and keep as baseline tone.
8. Adjust only below 500Hz or Schroeder frequency(where you think the room becomes modal) removing only peaks in the response with whatever Q filters you need to do the job - do not remove floor bounce.
9. Remeasure the speaker from the listening position with correction applied.
10. Use 1:1 smoothing response of measurement and raise it with broad filters to match the previous baseline 1:1 measurement.
Notes:
All of these steps should be performed for each speaker independently. The last step is probably the most overlooked and important.
A minidsp will probably run out of filters halfway through this method, so as you go there will need to be a refactoring of filters to reduce their number which attempting to maintain the effect.
Neumann and Grimm are two that immediately spring to mind.I do not know of many speakers that are able to achieve +/-1.5dB. Probably only Genelecs... any other ones?
I do not know of many speakers that are able to achieve +/-1.5dB. Probably only Genelecs... any other ones?
I seriously doubt you heard a 0.1dB difference, even if its broadband. That is an EXTREMELY minimal change.Pozz is absolutely correct. I have first hand experience of a couple of people noticing a 0.1dB broadband change. Could not identify what it was, but noticed that it was "different" than before. Of course they knew the replay system very well (a studio), and it was blind (they did not know anything had changed, it was not a "test").
I did not say I heard a 0.1dB difference... Read the post. I said a couple of people noticed it. I was the joker who changed the HF level in the studio by 0.1dB while they were out... Also read the literature, there are numerous papers and examples of people noticing a 0.1dB difference - Bob Katz said he could not believe his new DAC (a Prism or Craneson I think) sounded SO much better than his old one, more detail etc etc, until he checked levels properly, and found it 0.1dB hotter than his old DAC (probably an Avid). When he had the levels exactly the same he said he could not hear any difference between the DACs. No one hears a 0.1dB change as a level change, we notice it in other ways, "detail", "presence", "clarity" etc etc.I seriously doubt you heard a 0.1dB difference, even if its broadband. That is an EXTREMELY minimal change.
Every speaker designer I've ever talked to has assured me that their speakers measure flat. Flat can be defined in a multitude of ways, unless you are dealing with perfect constant directivity, a +-0db on-axis speaker will not be 'flat'. Currently regardless of who you purchase a pair of speakers from, or how much you paid, you can always do better correcting them yourself - if you know what you are doing.
Here are my 10 steps to achieve a "flat" speaker and room combo for one listening position:
1. All measurements for corrections are done windowed outside for treble on the tweeter axis
2. All bass measurements for corrections are done using 2 meter ground plane method outside - do not window this measurement down.
3. Splice the two measurements together and correct flat +-0.5dB to at least 50Hz
4. Remeasure off axis(0,5,10,15,30,45,60 degrees horizontal and 0,5,10,15,30 vertical) treble response outside with correction applied.
5. Average the results of the second round of measurements, any irregular dips or peaks that do not follow the slope in average treble should be adjusted minimally to within +-1.0dB of the natural directivity slope of the speaker, compromising the on-axis response for a smooth weighted power response.
-- You can stop at this point if you want a pure "flat speaker" and let the room do what it wants, otherwise continue --
6. Place corrected speaker in final position for best imaging, measure speaker with the mic pointed in listening position at ear level pointing at the speaker - do not window this measurement down.
7. Take 1:1 smoothing response and keep as baseline tone.
8. Adjust only below 500Hz or Schroeder frequency(where you think the room becomes modal) removing only peaks in the response with whatever Q filters you need to do the job - do not remove floor bounce.
9. Remeasure the speaker from the listening position with correction applied.
10. Use 1:1 smoothing response of measurement and raise it with broad filters to match the previous baseline 1:1 measurement.
Notes:
All of these steps should be performed for each speaker independently. The last step is probably the most overlooked and important.
A minidsp will probably run out of filters halfway through this method, so as you go there will need to be a refactoring of filters to reduce their number which attempting to maintain the effect.
But it seems to me like not a lot of bookshelf speakers (maybe none, actually) can be corrected to flat down to 50 Hz without bottoming out or struggling like crazy even at moderate volume, especially if you also give a boost thanks to that step 10.